Private Analytics Conversation 1.2

By Jim Sterne on Oct 18, 2020 in Articles by Jim Sterne

Analytics Cohorts meetings continue to be fascinating discussions about what's on people's minds at the moment, of the moment, and in the moment.

The first session covered a lot of ground like The Job of being a Digital Analyst, how to stop talking about data with the business side of the house, the tyranny of the ad-hoc question, and so much more.

Once again, the conversation covered a lot of topics and here's a summary.


Because I'd like to fire up a couple of more Analytics Cohorts and if you see value in participating in these conversations, then I'd like you to apply for a cohort right now.

But first:

What is this Exclusive Group?

Analytics Cohorts is a small, private conversation between consenting digital analysts. It's a safe space for talking about people, process, and sometimes technology. Careers, corporate relationships, fears, aspirations... it's a free-range exchange.


Split Implementation and Analysis?

Splitting functions
Is splitting functional responsibilities a good idea?

When told there was some consideration of moving their digital behavior capture team over into the Reporting group, this Senior Manager popped up their JIRA statistics to show that only about 40% of the output was actually creating reports. They didn't say, "No," right out of the gate because that's a card one can only play a couple of times in any organization before it's a career-stopper. This called for the improv idea of, "Yes, and..."

So, they reframed the conversation to about the people who put the data into the database - the implementation and pipeline people - and people who analyze the data to do things with it. Reporting is indeed one of those things, but the list is pretty long after that:

              Governance and access
              Marketing automation
              Product improvement
              Error detection
              Distribution for self-service
              A few other things
              And, finally, analysis

 But if you split off the implementation people from the analytics group, don't the analysts lose some of nitty-gritty, first-hand knowledge of the data itself? Isn't there harm in removing them from the context of the data and, in turn, does that diminish their deep understanding of the data?

While this is a serious concern, one person (or one team) cannot do everything. Some division of labor is necessary at some point. And when you do divide the teams, you can concentrate and maximize the skills needed to ensure good data - Grade A Data - and avoid garbage in and garbage out. This is a priority because a bad analysis can be reviewed and corrected and offers a learning moment. But bad data precludes any value coming out the other end. There is no recovery.

There are a few drawbacks to watch out for.


Analytics people earn their value by understanding their client audience. Trying to explain the complexity of how you analyze to the business side might be just the ticket for some clients and wasted on others. Some are fresh recruits who have never logged into an analytics tool. For them, page name, visits, and visitors are great metrics. Others have deep experience, are deep in the weeds, and understand the whole tech stack.

But a pure implementation team isn't working side-by-side with the business side of the house. When making technical decisions, they might not realize that their target audience may or may not be able to make use of the work they've spent weeks putting together. That disconnect must be addressed for this division of labor to be fruitful.

What's the Relationship Between Business Analytics, Customer Analytics, and Digital Analytics?

One of our Cohort members is just getting into digital analytics and wanted some guidance on what courses to take. There are so many to choose from and they have a great deal of overlap. What's the best place to start?

It starts with an overarching framework of how the pieces fit together.

Analysis/Analytics is a broad category. Learn some statistics to start with.

Business Intelligence looks at internal operations like shop floor control, and supply chain management.

Customer Analytics includes marketing, advertising, and customer service.

Digital Analytics is a part of Customer Analytics and includes:

Web analytics
Search analytics
Email analytics
Social analytics

These are not cleanly divided as the Call Center becomes more digital, Marketing Operations is different from Return on Marketing Spend, and Digital Analytics includes digital product analytics beyond just the user experience. But this taxonomy might help. Just remember that all models are wrong, but some are useful.*

Cookie Consent Implementation

Turns out cookie consent is quite a complicated project. A couple of our members were hoping for a better way.


Your legal team probably noticed GDPR at the last minute, or at least looked up when CCPA appeared on the horizon. So, you cobbled together a short-term solution with whatever your data tool vendors could help with, but now you know you need to start lashing together all of those disparate channels to bullet-proof consent.

The goal of a consent manager is to turn cookies on and off for marketing and for analytics, while leaving your essential cookies on. When your website uses a half dozen tracking tools (or more) and has other code that doesn't go through your main tag management system, things get sticky quickly.

Time to look to another set of tools like OneTrust, CookieHub, and Secure Privacy and be prepared to reach for the aspirin. They talk about having lots of integrations, but you still have to tweak all the code on your site. If your developers are setting cookies in 10 different places, you have to manually address each one and categorize everything on your page so all the toggle switches can work with all those categories. Frustrating work.

And it's not done yet.

Sequencing which tags should fire or not is based on conditions. You're likely to end up with a fragile tree of rules that have multiple firing conditions giving you additional conditions to each and every tag in your container. Or you can adjust the actual scripts that triggers consent. That means grabbing the value of the cookie - depending on which level of consent they have - and adding that variable into every tag. Oh, and make sure that the data layer fires before that consent manager because - timing. And did you account for different behaviors within the EU and where in the US where consent isn't required?

Ultimately, it's a ton of work, no matter how you bake it.

Professional Network

Professional Network

The conversation then swirled around job openings, job interviews, and who-do-you-know? That's always a wonderful conversation because we all know a lot of people on LinkedIn - but only on LinkedIn. When you can speak directly with people in your own Cohort, you find out there are often more connections than you thought. Six degrees of separation, anybody?

This time, the best advice when interviewing was to never agree to report to somebody who doesn't know what you actually do for a living. If they don't know analytics, they can't be in our corner when things get wonky.

Red Flag
The next-best advice was to ask your potential boss/team what they think the biggest challenges will be in your first 30 and 60 days, and what specific must-do tasks are lined up for the first month.

If they can't confidently answer those, it's a bit of a red flag. Just how strategic are they? The immediate tasks needs are way below your paygrade? Another red flag.

Finally, make a point of meeting the team you will be working with. It's nice to be part of a cohesive team with a holistic approach to analytics that you can collaborate with and learn from. That way, your know your work is important and you can make a real impact instead of feeling like you’re in a tech support roll. "Analytics Customer Service; How can I tweak your dashboard today?" Being part of a concerted effort toward a larger goal is central to job satisfaction.

2020: The Year of Bad Training Data

Some companies are barely hanging on and some are going great guns because this year is unlike anything we've ever seen and - hopefully - ever will again. This is a special circumstance and that makes it impossible to predict anything year-over-year. Every month is different. Heck, every week is different.

Forecast in COVID Times

So, start by comparing this year to last year just to see how different it is. What has changed? Then roll out all your usual tricks. Go to your stakeholders and say, "This is an interesting thing. What does it mean to you?" See what they come up with.

Look for the conversion pinch points. If you have increased interest but they're not necessarily well qualified, maybe you have a need for more targeting advertising, better descriptions on the website, more video content, a more robust FAQ, and/or a chatbot.

Then you can evaluate the value of the necessary investment, go to the team and say, "The numbers suggest that we need to answer these seven questions today on the website so that the call center gets a break, the email load lightens a little bit, and only qualified leads come in."

Always walk in with solutions. That will make you a hero and trusted advisor.



What is Analytics Cohorts?

Analytics Cohorts is a small, private conversation between consenting digital analysts. It's a safe space for talking about people, process, and sometimes technology. Careers, corporate relationships, fears, aspirations... it's a free-range exchange.

Apply today.



Photo by Simon Rae on Unsplash
Photo by John Barkiple on Unsplash
Photo by Дмитрий Хрусталев-Григорьев on Unsplash
Photo by Annika Ibels on Unsplash
Kitchen photo created by freepik


Sterne Measures Newsletter

Sign up and stay up to date on Jim's papers, articles, podcasts, and in-person appearances.

You'll receive a confirmation email because double-opt-in is in my blood.

Videos by Jim Sterne